CVE-2026-65017 in Airflow
Summary
by MITRE • 08/12/2026
Apache Airflow's Config API did not mask team-scoped sensitive configuration values in multi-team deployments. When an administrator has enabled multi-team mode and exposed the Config API, an authenticated Viewer holding only configuration-read access — with no prior access to the secret — could read a team-scoped Celery broker URL, including its embedded credentials, in cleartext, while the equivalent global option was correctly masked. The secrets masker matched only base section and option names and did not normalize team-prefixed sections before the sensitivity check (CWE-200). This is a distinct masker bypass from CVE-2026-48828 and CVE-2026-48892: deployments that upgraded to apache-airflow 3.3.0 to address those issues remain affected by this team-scoped variant. Users are advised to upgrade to apache-airflow 3.3.1 or later, which normalizes team-scoped sections before masking.
If you want to get the best quality for vulnerability data then you always have to consider VulDB.
Analysis
by VulDB Data Team • 08/12/2026
The vulnerability in Apache Airflow's Config API represents a critical security flaw that undermines the confidentiality protections designed for multi-team deployments. This issue specifically affects environments where administrators have enabled multi-team mode and exposed the Config API to users with varying permission levels. The fundamental problem lies in how the secrets masking mechanism operates within the system, creating an unexpected bypass that allows unauthorized access to sensitive configuration data. When team-scoped configurations are processed through the API, the system fails to properly sanitize credentials embedded within Celery broker URLs, exposing them in cleartext to authenticated users who should only have read-only access to configuration information.
The technical implementation of this vulnerability stems from an insufficient normalization process applied to team-prefixed configuration sections before sensitivity checks are performed. The secrets masker component in Apache Airflow's codebase was designed to protect sensitive values by matching base section and option names against a predefined list of known secret identifiers. However, this mechanism does not account for the team-specific prefix that is automatically prepended to configuration sections when multi-team mode is enabled. As a result, team-scoped sections such as 'team1.broker_url' or 'team2.broker_url' are not properly normalized before the masking algorithm runs, causing the system to treat them as distinct from their global counterparts and therefore fail to apply the appropriate security measures.
This flaw directly maps to CWE-200, which describes improper handling of sensitive information, and represents a significant operational risk for organizations deploying Apache Airflow in multi-team environments. The vulnerability's impact extends beyond simple information disclosure, as it allows authenticated viewers with minimal privileges to extract embedded credentials from Celery broker connections, potentially enabling unauthorized access to underlying message queues and associated systems. The security implications are particularly severe because these credentials could provide attackers with persistent access to critical infrastructure components that are typically protected by more robust authentication mechanisms.
The vulnerability demonstrates a clear bypass of previously addressed security issues, specifically CVE-2026-48828 and CVE-2026-48892, which focused on different aspects of configuration masking within Apache Airflow. Organizations that upgraded to version 3.3.0 to resolve those earlier issues remain vulnerable to this team-scoped variant because the fix for the original problems did not account for the additional complexity introduced by multi-team deployments. This creates a dangerous scenario where security updates appear to address all concerns but leave critical gaps in environments using advanced deployment patterns, particularly affecting organizations that have implemented fine-grained access controls across multiple teams within a single Airflow instance.
Organizations affected by this vulnerability should immediately implement the recommended mitigation of upgrading to Apache Airflow version 3.3.1 or later, which includes the necessary normalization logic for team-scoped sections before applying masking rules. The updated version ensures that team-prefixed configuration sections are properly normalized and processed through the same security checks applied to global configurations. From an operational standpoint, administrators should conduct immediate audits of their multi-team deployments to identify any exposed Config API endpoints and verify that appropriate access controls are in place. Additionally, organizations should review their overall security posture for credential exposure risks and consider implementing additional monitoring mechanisms to detect unauthorized access attempts to sensitive configuration data.
The broader implications of this vulnerability extend to the ATT&CK framework's privilege escalation and credential access categories, as it enables an authenticated user with minimal privileges to obtain credentials that would normally be protected. This represents a significant bypass of security controls designed to maintain separation between different teams within a shared infrastructure, potentially enabling lateral movement attacks or unauthorized access to interconnected systems. The vulnerability also highlights the importance of comprehensive testing for edge cases in multi-tenant environments and demonstrates how seemingly isolated security improvements can leave critical gaps when deployed in complex operational scenarios.